Todays result: a camera mount

I have been trying to take night pictures lit up cities with a 15 second exposure time but I struggle to keep my camera perfectly stable. The best vantage locations are high up and usually a little windy. I have a small tri pod but it is easily blown around and results in a blurry image. 

I have printed a simple camera mount with a wide base for stability that holds the camera flat. It is not adjustable so to angle it I need to find a appropriate surface or place things under it. It is wide enough to place something heavy on the base to hold it down if need be.

The base is designed to print at 0.3 mm layer height, and the screw is designed to print at 0.1 layer height. Both models include an anchoring base layer to avoid warping. The screw model has a cooling structure around the part. I was successful printing in ABS at 75mm/s for the base and 20mm/s for the screw.  After printing run a knife around the edge of the parts to remove the construction material. 

The models are avaliable here

Assembled with camera

Picture of Adelaide

Blurry picture of Adelaide,

[note the trail through the sky left by a plane]

Parts

Close up of 1/4" thread

Todays result: Combinatory pixel based image processing

I have made some simple image processing software that works by applying combinations of functions to pixels in the image. Functions fall into a group determined by their input type and calculated output type. The program systematically combines functions into all permutations and applies them to the image. I have used every double <- double function in the C# math library. This does not really make sense, but some of the nonsensical combinations make interesting results. The program doesn’t really produce very meaningful results but some of them look nice.

Function categories

• pixel <- value  
• value <- pixel
• value <- value
• pixel <- pixel
• image <- image

An example combination

1. Function 1: Calculate the sum of elements in a pixel
2. Function 2: Calculate the mean and standard deviation of all (Function 1) pixel values, and then calculate the Z values for each pixel
3. Function 2: Map the Z values back to pixels by interpolating between black where z = -3, and white where z = +3.

Results of processed images

Some more results of processed images

Todays result: 3D cellular fill patterns

3D cellular structures should be the standard fill mechanism used by slicing software. 

My recent work has lead me to the conclusion that 3D cellular structures make better fill patterns for 3D printing and that it would fit neatly as the fill pattern used by slicing software. Using cellular fill will have benefits over using horizontal alternating diagonal fills, and over 2D pattern fills such as hexagonal tessellations. 

Using cellular fill increases the strength of the part relative to the amount of material. Conversely, less material is needed to achieve the same overall strength. The cellular fill is a deliberate structure with plastic used sparingly but in a focused way. Each cell is a pocket of empty space with only the outer perimeter traced in plastic. This approximates a sphere producing a strong shape relative to the amount of material used. Each cell shares material with, and benefits from the stability and strength imparted by other cells. Using less material leads to less printing time.

This work has been my attempt to contribute back to the Reprap community. I have produced proof of concept model using the Pink panther woman but I feel pressured to release this idea before I can get it fully operational because of the recent plague of 3D patents and trolls.

In the near future, the use of cellular fill structures will be a normal and automatic part of slicing software instead of the current zig-zag fill a cellular pattern will used. At each layer the slicing engine will also slice (or more likely compute) the pattern and insert it into the object.

The generation of the pathways by slicing engines will also need to change. When sliced horizontally an object filled with cellular structure looks like a collection of different sized circles compressed together.  For the best outcome, every cell needs to be constructed as a series of complete circuits around the perimeter of the cell. In my experience, current slicing software tends to identify some cells as complete circuits, while others are afterthoughts once the complete cells are in place. The result is poor with the regions of the afterthought cells not connecting correctly to one or both adjacent walls. To date experimentation has been using objects filled with cellular structures within the model itself. 

Similar to the selecting layer height and print speed, the user will select the pattern used, wall thickness and size of the pattern. Orientation and location of the pattern may be advanced options.  By varying the settings the desired strength, density, and timing trade-offs can be controlled.

The main pattern of use is the truncated octahedron. It is a beautiful shape, made of six squares and eight hexagons all with equal length sides. It is single shape that will completely occupy 3d space with no gaps. It tessellates through uniformly via translation alone ie no rotations or other transforms. I have generated other more complicated patterns that could be used but I don’t believe at this point that they offer any additional advantages.

 Truncated octahedron pattern

This print is 100 * 100 * 50 mm and weight ~50gm.

It will support over 80kg placed on the thinner edge
I estimate it uses approximately the same quantity of plastic as a 10% zig zag fill

The Pink panther woman, scaled larger and cut into six segments.

Each segment is filled with a different cellular pattern

Printed in transparent ABS to show off the internal structure

Individual segments can be seen in previous posts.

The truncated octahedron model will be made available soon.

Todays result: A box

I have made a box to cover my printer. It is made out a fruit packing box sourced from my local supermarket. It is a 65cm cube of 15mm thick cardboard and supports my full weight from the top. 

The original box carried pumpkins. It was a large octagon shape about 1.8m across with 4 square panels alternated with four smaller rectangular panels. Each panel had a flap that was stapled to a wood pallet.  I cut the box into pairs of one squares and one rectangle. The sections where reassembled into a cube with one face missing. Fencing wire and small MDF blocks where used to stitch the panels together. The wire was pushed through the cardboard and looped through a piece of MDF on either side. I sealed the edges and the holes left by handles in the original box.

I made to box to help keep it the printer warm while printing ABS. It has the added benefits of containing plastic smells and dampening some noise. 

The printer cables travel under the edge of the box.  I keep my circuit board outside of the box for better cooling. The MOSFETs get very hot and the steeper drivers warm a little. The box puts too much pressure on the Z-motor cable which is the shortest connection to the printer. The circuit board connector bends slightly and every time I lift the box to check progress I fear something will go wrong. 

 Box, grey sealer applied liberally, and fasteners visible

Model before assembling box to ensure it would fit.

Todays result: A full set of Mendel90 parts

I now have one or more spares for every printed Mendel90 part.

I have lost count of the number of parts on my original Mendel Prusa style printer that I have replaced. It now is made of more replacement printed and electronic parts than original parts. After a period of being broken, once operational again the first thing I did was to print a spare for every printed part. 

The reasons parts have broken

• PLA parts warped when printing ABS,
• Extruder gears broke from ware,
• Parts broke from stress or over tightening,
• Generally delicate parts - the manufacture appears to have reduced the printed volume of many parts.

The same applies to the Mendel90, I have had to make at least four repairs now. The broken parts have either been PLA warping at ABS temperatures or bad prints to begin with. The Mendel90 is a much more reliable design. 

Having a 'spare' printer has reduced the need to backup the Mendel90, but I am also keen to one day convert the Prusa style printer into a Mendel90. I have been using periods between large prints to print a spare or two. The majority have been in PLA. I got carried away trying to print everything quickly and used PLA for some items that need to be ABS. I now have a few PLA spare of spares.

When printing ABS or for long periods in an enclosed chamber, parts near the extruder or in contact with the X Y or E stepper warp if made from PLA. PLA seams fine for the other parts.

In my experience parts that should be ABS

• x-motor 
• y-motor bracket
• d_motor_bracket
• wades_block
• wades_gear_spacer
• wades_idler_block
• x_belt_clamp
• x_belt_grip
• x_belt_tensioner
• x_carriage
• x_carriage_fan_bracket
• x_carriage_fan_duct
• x_motor_bracket
• y_motor_bracket

Full set of Mendel90 v2 printed parts (grey PLA, black ABS)

Mendel Prusa style printer, Note that the original part colour was dark grey.

Todays result: Carbon filter

I have created a filter to add to my printer. It helps by removing some of the bad plastic odours. The filter medium is activated carbon pellets purchased at an aquarium shop.

It is a box shape with a fan at both ends and a diagonal wall of carbon through the middle. The intent was to double the force moving the air through a large area of thin carbon. It was designed to mount on the side of my Mendel90 it fits neatly on the inside of the right stay. I directed the air flow downwards. When the printer is operating in enclosed box it will pull warm air back down from above.

The y length is 17cm, almost the full length of the printer bed. The z height is 70mm - the fan size, and the x width is fan size + 2 cm wall of carbon. The filter grill is spaced with 3mm gaps to hold the 3.6mm diameter pellets.  I used 70mm fans due to a surplus from a botched order and 2cm of carbon seamed a useable thickness. Each end has m3 nut traps to attach the fan, and a slide cover to hold the carbon. 

The very bottom layer is mostly missing. The second layer droops a little when printed and restores a flat bottom. All but a few mm around the perimeter of the bottom layer have been removed. This reduces the adhesion to the bed otherwise it would be very difficult to remove.  To avoid severe warping the bed needs to be heated. I used PLA, ABS is a different beast where a complete bottom layer would probably work better.

The top came out well; I have to admit I was surprised, but it is not air tight. Along the sides of the top there are gaps between strands of plastic. This is where the largest bridging was. I suspect in 1 or 2 more layers it would have completely sealed. Painting it should resolve this. The model has inbuilt support structures and does not need generated support when slicing. The supports for the top branch off the filter grill and walls. They snap away easily after printing allowing access to the nut traps. The nut traps are still fiddly to use as they face into the chamber.

This is my third filter attempt; the previous filters had too much carbon and almost no air flow. 

The model is available here, though I am not releasing the scad files due to objections to thingiverse. I may produce stl files for other configurations if you leave a request it in the blog comments (no promises!).

 Mounted filter box on Mendel90

Failed filter attempt #2, converted broken PC power supply box. 

Electrical tape covered the screw holes and a tissue was placed inside the box over the vents on the far side.

Failed filter #1, printed mesh tube ends, designed to hang off my spool holder rod.